This document presents a new iris segmentation method for iris recognition systems. The proposed method uses Canny edge detection and Hough transform to locate the iris boundary after finding the pupil boundary using image gray levels. Experiments on the CASIA iris image database of 756 images show the method can accurately detect the iris boundary in 99.2% of images. This is an improvement over other existing segmentation techniques. The key steps of the proposed method are preprocessing, segmentation using Canny edge detection and Hough transform, normalization using the rubber sheet model, feature encoding with Gabor wavelets, and matching with Hamming distance.